University of Pretoria
Browse

Investigation of the performance and interpretability of two models, a large language models (LLM) and a small-scale model, trained on low-resource language pairs Xhosa Zulu and Tswana-Zulu

dataset
posted on 2025-01-22, 09:17 authored by Tsholofelo GombaTsholofelo Gomba

This submission contains images and datasets used in the research for a dissertation "Assessing interpretability in machine translation models for low-resource languages".

The images include machine translation model-generated heatmaps and machine translation model-generated translations.

The datasets include the following:

  1. BLEU scores from model training and graphs
  2. Post model evaluation results for MQM and graphs
  3. Post model evaluation results for ESS and results
  4. Small-scale model training results comparisons with generated graphs [to evaluate early stopping]


History

Department/Unit

Engineering, Built Environment and Information Technology

Sustainable Development Goals

  • 4 Quality Education
  • 9 Industry, Innovation and Infrastructure